Loss of VLDL Receptor Activates Retinal Vascular Endothelial Cells and Promotes Angiogenesis

Abstract

purpose. The very low-density lipoprotein receptor (VLDLR) knockout (vldlr ^−/−) mouse has been identified as a model for retinal angiomatous proliferation with subretinal neovascularization (SNV) evolving from retinal vessels. The effects of VLDLR on the angiogenic functions of retinal vascular endothelial cells (RVECs) in vivo and in vitro were examined.

methods. Immunofluorescent staining of markers for activated endothelial cells was performed with CD105 and CD106 antibodies. Proliferation, tube formation, and migration assays were carried out in RVECs isolated from wild-type and vldlr ^−/− mice to assess the angiogenic functions in vitro. The effect of VLDLR blockage on wild-type RVEC proliferation was also examined.

results. The expression of CD105 and CD106 was significantly upregulated in the retinas of adult vldlr ^−/− mice, especially at lesion sites. An intense CD105 signal was found in the inner retinas of vldlr ^−/− mice starting at postnatal day 14, before the onset of SNV. In vitro proliferation assays revealed a significantly enhanced (approximately 20%–100%) growth rate in vldlr ^−/− RVECs compared with that in the wild-type RVECs. The formation of capillary-like structures in vldlr ^−/− RVECs was approximately 3 to 11 times greater than in wild-type RVECs. Migration of vldlr ^−/− RVECs was 1.3 to 3.7 times that of wild-type. VLDLR blockage using a receptor-associated protein or neutralizing anti-VLDLR antibodies significantly enhanced the proliferation rate in wild-type RVECs by more than 200% and 30%, respectively.

conclusions. VLDLR is a potent endogenous inhibitor that negatively regulates the angiogenic properties of RVECs. Loss of VLDLR activates RVECs and significantly enhances angiogenesis in vivo and in vitro.

Age-related macular degeneration (AMD) is the leading cause of severe vision loss in elderly persons. ¹ Although most cases of neovascular AMD are caused by choroidal neovascularization (CNV), a subgroup of neovascular AMD, known as retinal angiomatous proliferation (RAP), is characterized by subretinal neovascularization (SNV) arising from retinal vessels. RAP, representing approximately 12% to 15% of patients with newly diagnosed neovascular AMD, has its own clinical course and prognostic parameters that differ distinctly from other forms of neovascular AMD. ^{2 3 4 5 6} In patients with diagnoses of unilateral RAP lesions, the form of neovascularization that develops in the fellow eye is virtually always RAP. Annual and accumulative risks of neovascularization in the fellow eye are higher in patients with RAP than in those with other forms of neovascular AMD. ⁷ The response of RAP to photodynamic therapy is different from that of classic CNV. ² These studies suggest that SNV and CNV develop from different etiologies and with different mechanisms. 

Recently, VLDLR has been identified as one of the functional candidate genes for a significant association with AMD in humans. ⁸ VLDLR is an 86-kDa transmembrane protein that belongs to the LDL receptor family. Remarkable interspecies conservation (>95%) of the VLDL receptor cDNA sequence among human, ^{9 10 11 12} mouse, ^{13 14} rabbit, ¹⁵ and rat ¹⁶ suggests an important physiological role for this receptor. The VLDLR is most abundantly expressed in heart, skeletal muscle, and adipose tissue, ^{9 12 13 14 17} localized in endothelial and smooth muscle cells of blood vessels. ¹⁸ VLDLR binds to numerous ligands, including apolipoprotein lipase (LPL), receptor-associated protein, thrombospondin-1, urokinase plasminogen activator (uPA)/plasminogen activator inhibitor-1 complex, and several other proteinase-serpin complexes. However, the function of VLDLR in the retina and in retinal angiogenesis remains unknown. 

The discovery of consistent SNV in a germ line knockout mouse of the gene encoding VLDLR provides a reproducible animal model in which to study the role of VLDLR in retinal neovascularization. ^{19 20} Complete penetration of the retinal phenotype in the vldlr ^−/− mouse indicates a strong association between retinal neovascularization and the VLDLR mutation. It also suggests a prominent inhibitory effect of VLDLR on retinal angiogenesis. We and others have shown that vldlr ^−/− mice exhibit histologic and angiographic characteristics of RAP with age-related SNV evolving from retina. ^{21 22} Our previous study also mapped VLDLR expression in the retina, particularly in retinal vascular endothelial cells (RVECs). ²² However, little is known about the effects of VLDLR on retinal endothelial cells. 

Endothelial cells play a pivotal role in angiogenesis and are considered to be ideal therapeutic targets for inhibiting angiogenesis. ²³ At resting/quiescent conditions, they are intimately involved in maintaining the blood-tissue interface. In many pathologic conditions, including inflammation and angiogenesis, endothelial cells can be activated with the expression of several proangiogenic molecules and activation markers, such as CD105 (Endoglin), CD106 (vascular cell adhesion molecule-1), CD141, and platelet activating factor. ²⁴ Endothelial cell activation is involved in multiple phases of the angiogenic process, including initiation, prolongation, and exacerbation of angiogenesis. To determine the role of VLDLR in retinal endothelial cell activation, we tested the hypothesis that VLDLR is a critical molecule for maintaining the quiescence of RVECs, and loss of VLDLR in vldlr ^−/− mice resulted in RVEC activation and uncontrolled vascular overgrowth leading to retinal neovascularization. In the present study, we compared the activation of endothelial cells in the retina of wild-type and vldlr ^−/− mice by examining the expression of CD105 and CD106. Angiogenic functions of RVECs isolated from both genotypes were also studied. 

Methods

Experimental Animals

All animals used in the study were maintained and treated with strict adherence to the guidelines for animal care and experimentation as set forth in the ARVO Statement for the Use of Animals in Ophthalmologic and Vision Research, and all procedures were approved by the Indiana University Animal Care and Use Committee. Breeding pairs of mutant mice with targeted deletion of the VLDLR gene (B6; 129S7-Vldlr^tm1Her/J; vldlr ^−/−) ²⁰ were obtained from the Jackson Laboratory (Bar Harbor, ME). Breeding pairs of a transgenic line expressing a temperature-sensitive SV40 large T antigen were obtained from the Charles River Laboratories (ImmortoMouse; Wilmington, MA). Wild-type (C57BL/6J; +/+) and vldlr ^−/− mice were crossbred with the transgenic mice. All offspring were subjected to PCR genotype confirmation of SV40, wild-type VLDLR, and vldlr knockout genes. Cells isolated from the littermates produced from these crosses were immortalized by turning on the SV40 gene in vitro and were used for all in vitro studies. 

Immunofluorescent Staining

Eyes of wild-type and vldlr ^−/− mice at ages of 2 weeks, 3 weeks, and 1 month were embedded in OCT compound (Miles Inc., Elkhart, IN) and immediately frozen at −80°C. Radial sections of 12-μm thickness were cut at −20°C. After they were fixed in 1% paraformaldehyde (PFA), sections were blocked with a signal enhancer solution (Invitrogen, Carlsbad, CA) at room temperature for 30 minutes. Sections were incubated with a rat anti-mouse CD105 (1:100) or CD106 (1:100) antibody (eBiosciences, San Diego, CA) at 4°C overnight. After they were washed in PBS, sections were incubated with an FITC-labeled secondary antibody. Cell nuclei were counterstained with DAPI. Negative controls were included by omitting the primary antibody to confirm the specificity of the antibodies. 

RVEC Primary Culture

Mouse RVECs were isolated as described previously, with modifications. ²⁵ Briefly, retinal tissues from young adult wild-type and vldlr ^−/− mice carrying a SV40 transgene were digested in DMEM with 0.2 mg/mL collagenase type I (Worthington, Lakewood, NJ) at 37°C for 3 hours. After trituration and filtration, the dissociated cells were incubated with sheep anti-rat magnetic beads (Dynal Biotech, Lake Success, NY) precoated with a rat anti-mouse PECAM-1 monoclonal antibody (BD Biosciences, San Jose, CA) for affinity binding. Bead-bound cells were plated in a 24-well plate precoated with 2 μg/mL of human fibronectin (BD Biosciences) in an endothelial cell growth medium (EBM/EGM-2MV; Cambrex, East Rutherford, NJ). RVECs were maintained in culture at 33°C with interferon to turn on the SV40 gene, but kept under ordinary conditions at 37°C in the absence of interferon in all assays. 

Immunocytochemistry

RVECs cultured in an 8-chamber slide (Electron Microscopy Sciences, Plymouth Meeting, PA) were fixed, blocked, and incubated with various primary antibodies including a rat anti-mouse CD31 antibody (1:100; BD Biosciences), a goat anti-mouse VE-cadherin antibody (1:100; Santa Cruz Biotechnology, Santa Cruz, CA), and an anti-αSMA-FITC monoclonal antibody (1:1000; Sigma, St. Louis, MO) at 4°C overnight. After thorough washing in PBS, sections were incubated with a corresponding fluorescence-labeled secondary antibody. Cell nuclei were counterstained with DAPI. 

Cell Proliferation Assay

RVECs were starved in basal medium (BM; EBM supplemented with 0.5%FBS) at 37°C overnight. After trypsinization, cells were resuspended in BM or BM plus various reagents and then were seeded at 2000 cells/well in 96-well plates. For VLDLR blocking experiments, a series of doses of recombinant human or rat receptor-associated protein (Innovative Research, Southfield, MI) was used. Neutralizing anti-VLDLR antibodies (Santa Cruz Biotechnology) and a group of control IgG (Sigma) were included. After 48 hours incubation at 37°C, the total number of cells was quantified (CellTiter 96 AQueous One Solution; Promega, Madison, WI). The absorbance value at 490 nm was read by a microplate reader (GENios Pro; Tecan Trading AG, Geneva, Switzerland) after 1-hour incubation. Experiments were performed in five to six wells per group and were repeated three times. To generate a growth curve, wild-type and vldlr ^−/− RVECs were resuspended in BM or growth medium (GM) and seeded at 500 cells/well in 96-well plates, respectively. On the following 5 days, the medium was changed every other day, and MTS assay was performed in quadruplicate. 

Tube Formation Assay

The formation of capillary-like structures by RVECs in growth factor-reduced basement membrane matrix (Matrigel; BD Biosciences) was performed as previously described with modifications. ²⁶ Wild-type and vldlr ^−/− RVECs at the same level of confluence were starved in BM for 24 hours before seeded at a density of 2 × 10⁵ cells/well in a 24-well plate precoated with 150 μL basement membrane matrix (Matrigel; BD Biosciences). Cells were incubated in BM, BM plus 10 ng/mL bFGF, or GM, at 37°C for 16 hours. After staining (CellTracker Green BODIPY; Invitrogen), images of six randomly chosen fields from each well were captured under an inverted fluorescence phase-contrast microscope using a 10× objective (DM IRB; Leica Microsystems, Bannockburn, IL). The lengths of the tubes in each field were measured with advanced digital imaging software (SPOT; Diagnostic Instruments, Sterling Heights, MI). The experiment was performed in triplicate. 

Migration Assay

Migration was performed using modified Boyden chambers containing polycarbonate membrane (Transwell, 8.0 μm pore size; Costar, Cambridge, MA) with slight modification from the previously described method. ²⁷ Wild-type and vldlr ^−/− RVECs at the same confluence were starved with BM overnight and then seeded at 5 × 10³ cells/well into polycarbonate membrane plates (Transwell; Costar) coated with fibronectin (10 μg/mL). BM, BM with 10 ng/mL bFGF, or GM was added to the lower chamber and incubated at 37°C for 16 hours. Viable cells were stained with a reagent (CellTracker Green BODIPY; Invitrogen) and counterstained with DAPI. The upper surface of the transwell member was scraped with a cotton swab to remove unmigrating cells. Six images of migrating cells were taken randomly from each membrane with a 40× objective. The average number of nuclei in each image was determined. The experiment was performed in triplicate. 

Statistical Analysis

All statistical analyses were performed with commercial software (SPSS 11.0; SPSS, Chicago, IL). Data are presented as mean ± SD. Differences were assessed using one-way or two way analysis of variance (ANOVA) or the Student’s t-test. P < 0.05 was considered statistically significant. 

Results

Endothelial Cell Activation in the vldlr ^−/− Retina

Endothelial cell activation is essential for several vascular pathologic processes, which can lead to inflammation and angiogenesis. To determine whether endothelial activation is involved in SNV in the vldlr ^−/− retina, immunofluorescence staining was performed to detect the expression of CD105 and CD106, markers for endothelial cell activation, in the retinas of wild-type and vldlr ^−/− mice. Control experiments were carried out without the primary antibody. The absence of immunoreactivity in control retinal sections confirmed the specificity of each antibody (image not shown). We found that the signal of CD105 immunoreactivity was minimal in the retinas of young adult wild-type mice (Fig. 1A) . At higher magnification, scattered dim circular, semicircular, and patchy stains were barely visible in the inner retina within the outer plexiform layer (OPL), inner plexiform layer (IPL), and ganglion cell layer (GCL) (Fig. 1B) . The distribution of the staining pattern corresponded with the retinal vascular structures revealed by lectin staining (Fig. 1C)in these layers. Only a small fraction of the vascular network was CD105 positive in retinal sections of wild-type mice. In contrast, CD105 signals in young adult vldlr ^−/− mice were much stronger throughout the entire retina than they were in the wild-type retina (Fig. 1D) . The strongest signal was detected at the neovascular bulb in the subretinal space (Figs. 1D 1E) . Striking RVEC activation was evident at postnatal day 14 (P14), and the onset was earlier than SNC in the vldlr ^−/− retina (Figs. 1G 1H) . CD105 staining in the P14 wild-type retina was weak, similar to that in the young adult wild-type (data not shown). The close match of the staining pattern between CD105 antibody and lectin (Figs. 1D 1F 1G 1I)in serial sections of the vldlr ^−/− retina indicated that the retinal vasculature was activated in its entirety, not just at the lesion site. A similar lectin staining pattern in the inner retina of wild-type and vldlr ^−/− mice implied that the density of retinal vessels was comparable between the two genotype groups except for the SNV. The difference in CD105 staining intensity and pattern implied the significant upregulation of CD105 in the vldlr ^−/− retina. 

Staining of the other endothelial cell activation marker, CD106, also revealed strong signals in the vldlr ^−/− retina (Fig. 2B) . However, the pattern was different from that of CD105. CD106 only stained structures such as large vessels in the wild-type GCL (Fig. 2A) . No signal was found in the capillary network in the IPL or OPL. In addition to the staining of large vessels in the GCL, an intense CD106 signal was found at the lesion site in the vldlr ^−/− retina marking the entire track of neovascular growth, which originated from the GCL and extended through the retina into the subretinal space (Fig. 2B) . Unlike CD105, there was no positive staining of CD106 in the vldlr ^−/− retina before the onset of SNV at P14 (data not shown). Strong staining of both markers in the vldlr ^−/− retina revealed continuous endothelial cell activation starting before the onset of SNV. 

Characterization of RVECs from Wild-Type and vldlr ^−/− Mice

Continuous endothelial cell activation can increase angiogenesis. To determine whether the endothelial activation in the vldlr ^−/− retina is associated with accelerated angiogenic functions, we used in vitro approaches to assay multiple aspects of angiogenesis. The method of isolation of murine retinal endothelial cells from the transgenic mouse (ImmortoMouse; Charles River Laboratories) has been described in detail. ²⁵ The combination of isolating RVECs with antibody-labeled beads and introducing the SV40 transgene to immortalize the cells enabled successful establishment of RVEC culture from the mouse retina. As a precautionary measure, animals after at least five generations of cross-breeding were used to minimize any potential effects of genetic background variation. The development of SNV phenotype was also confirmed in the transgenic vldlr ^−/− retina by lectin staining (data not shown). With this approach, we were able to isolate a homogeneous population of RVECs from wild-type and vldlr ^−/− mice. Figure 3shows a cobblestone-like morphology typical of endothelial cells from wild-type and vldlr ^−/− mice (Figs. 3A 3B) . 

To confirm that these cells were endothelial cells, we examined the expression of two endothelial cell specific markers, CD31 and VE-cadherin. ²⁸ As shown in Figures 3C and 3D , more than 95% of wild-type and vldlr ^−/− RVECs were stained positively for CD31 and VE-cadherin. An anti-αSMA antibody, a marker for pericytes, was used as a negative control in a double labeling experiment with CD31. Only occasionally αSMA-positive cells were seen in the RVEC culture (Fig. 3C , arrow), further confirming the minimal contamination of other cell types. The morphology and expression of the endothelial markers were maintained throughout the culture time used for all in vitro assays between passages 6 and 14. 

Significantly Faster Growth of vldlr ^−/− RVECs

In an initial RVEC culture, we noticed that RVECs isolated from vldlr ^−/− mice reached the confluent stage more quickly than those from wild-type mice. As shown in Figures 3A and 3B , vldlr ^−/− RVECs were almost confluent after 5 days in vitro, whereas wild-type RVECs were only approximately 50% confluent, indicating a rapid growing trend of vldlr ^−/− RVECs. Proliferation is an important index of angiogenic ability of endothelial cells. To quantify the growth rate, an MTS assay was used to determine the growth curve of RVECs from both genotypes over a 5-day growth period. The mean optical density (OD) reading of viable cells measured 3 hours after seeding in day zero in each group served as the basal point of 100%. As shown in Figure 4A , vldlr ^−/− RVECs grew consistently more quickly than wild-type RVECs in BM. The difference was noticeable from day 1 with an overwhelming statistical significance (n = 4; P < 0.01 by two-way ANOVA). The growth of RVECs reached a plateau on day 3 in both genotypes and started a descending trend on day 5 indicating potential apoptosis. When cultured in GM, growth curves of wild-type and vldlr ^−/− RVECs were boosted to a much higher level and lasted longer. The genotypic difference was further enhanced (n = 4; P < 0.001 by two-way ANOVA). Because GM contains several growth factors, including bFGF, VEGF, IGF, and EGF, the bigger genetic difference in the GM also indicated that vldlr ^−/− RVECs were more sensitive to growth factor stimulation. 

Enhanced In Vitro Angiogenesis in vldlr ^−/− RVECs

Tube formation assay is a well-defined in vitro angiogenesis assay to quantitatively evaluate the unique intrinsic property of endothelial cells, forming capillary structures in the basement membrane matrix (Matrigel; BD Biosciences). In this study, RVECs of wild-type and vldlr ^−/− mice were subjected to tube formation in three culture conditions—basal, bFGF (10 ng/mL) supplemented, and GM. Under the same seeding density, more extensive capillary formation was found in vldlr ^−/− RVECs than that in wild-type RVECs (data not shown). Quantitative measurement of total tube length in each well revealed a striking upsurge of capillary formation in vldlr ^−/− RVECs in all three medium (Fig. 4B) . While both bFGF and GM promoted an increased length of tube formation in wild-type RVECs, the genotypic difference was consistently present in all three culture conditions, with an approximately 3- to 11-fold increase in vldlr ^−/− RVECs than in wild-type RVECs (P < 0.01). 

Increased Trend of vldlr ^−/− RVEC Migration

Another important aspect of activated endothelial cells is migration toward angiogenic stimuli. To assess the ability for migration, wild-type and vldlr ^−/− RVECs were seeded in the upper chambers of 8-μm pore transwells and exposed to basal, bFGF-supplemented (10 ng/mL), or GM in the lower chamber. Quantification of the total number of migrating cells revealed significantly (P < 0.01) more migration of vldlr ^−/− RVECs than of the wild-type in BM and 10 ng/mL bFGF (Fig. 4C) . The number of migrating vldlr ^−/− cells was 1.3 to 3.7 times that of wild-type. An increased trend of vldlr ^−/− RVEC migration in the GM was also visible, but the increase was not statistically significant (P = 0.088). 

Blocking of VLDLR and Wild-Type RVEC Proliferation In Vitro

The enhanced angiogenic function of vldlr ^−/− RVECs over such a broad spectrum clearly indicates that genetic deletion of VLDLR promotes angiogenesis. To directly test the hypothesis that loss of VLDLR accelerates retinal endothelial cell growth, we examined the effect of VLDLR blocking on wild-type RVEC proliferation in vitro. Wild-type RVECs were treated with or without recombinant human receptor-associated protein (rhRAP), a VLDLR antagonist, at different concentrations ranging from 0.4 to 1.6 μM. A clear dose-dependent enhancement of wild-type RVEC proliferation was evident after rhRAP treatment (Fig. 5A) . Significant boosting effects (P < 0.001) on RVEC proliferation were seen when the antagonist dosage reached 0.8 μM (>100% increase) and 1.6 μM (>200% increase). Similar boosting effects of blocking VLDLR on proliferation were also observed when wild-type RVECs were treated with recombinant rat receptor-associated protein (rrRAP; Fig. 5A ). Because receptor-associated protein is a nonspecific antagonist to the LDL family that may also affect other members of the family expressed in these cells, a separate assay was carried out using a cocktail of anti-VLDLR neutralizing antibodies. Anti-VLDLR antibodies also accelerated the proliferation of wild-type RVECs by more than 30% (P < 0.01) compared with a nonspecific IgG control treatment (Fig. 5B) . These results further confirm an inhibitory role of VLDLR in regulating RVEC growth. 

Discussion

In this study, RVEC activation in the vldlr ^−/− retina was demonstrated by elevated expression of the endothelial cell activation markers CD105 and CD106 in immunohistochemistry. Activation was not restricted to SNV lesion sites but also appeared in the inner retina and was detected even before the onset of SNV at P14. In vitro functional assays of multiple angiogenic properties revealed that RVECs isolated from the vldlr ^−/− mouse retina had a significant upsurge of the growth curve, formation of capillary-like structures, and a trend toward increased migration when compared with those from the wild-type. Blocking of VLDLR with an antagonist, receptor-associated protein, or neutralizing antibodies significantly accelerated the proliferation rate of wild-type RVECs. These results demonstrate that VLDLR is a potent inhibitory molecule for RVEC angiogenic function. Loss of VLDLR activates RVECs and significantly enhances a spectrum of angiogenic properties of RVECs in vivo and in vitro. Our data support the hypothesis that VLDLR is a critical endogenous molecule maintaining the homeostasis of RVECs and that loss of VLDLR in vldlr ^−/− mice results in RVEC activation and induces uncontrolled vascular outgrowth leading to SNV. 

Potential Role of VLDLR in Maintaining Endothelial Cells in a Quiescent State

VLDLR is known to be expressed in endothelial cells of capillaries and arterioles of many organs, including skeletal muscle, heart, ovary, liver, and brain. ²⁹ The primary function of this receptor is considered related to the delivery of lipoproteins to cells. To date, no report directly links VLDLR to the quiescence of endothelial cells. However, a recent study showed that thrombospondins, presumably TSP1 and TSP2, acted through VLDLR to inhibit cell division in human neonatal dermal microvascular endothelial cells (HMVECs). ³⁰ The authors propose that TSP1 and TSP2, together with VLDLR, initiate a pathway for the maintenance of normal vascular endothelium in a quiescent state. Nevertheless, the exact role of VLDLR in endothelial cell quiescence is obscured by their findings that receptor-associated protein and VLDLR antibodies inhibited VEGF-stimulated increased HMVEC proliferation. In the present study, we found significant activation of endothelial cells in the vldlr ^−/− retina marked by the upregulation of CD105 and CD106. In addition, a series of in vitro assays revealed increased proliferation, tube formation, and migration of vldlr ^−/− RVECs. Wild-type RVECs could be activated, and their proliferation was accelerated by blocking VLDLR with its antagonist or with neutralizing antibodies, mimicking the increased proliferation seen in vldlr ^−/− RVECs. These findings suggest a potential role of VLDLR in maintaining retinal endothelial cells in a normal quiescent state. Although the reason for the discrepancy of VLDLR blocking effects on proliferation between the TSP study and our results is unclear, one possible explanation could be the different cellular origin of endothelial cells. For example, the response to TSP2 in HMVECs is different from that in human umbilical vein endothelial cells (HUVECs). ³⁰ It is well known that endothelial cells in different organs behave very differently under different conditions. This includes differences in expression levels of VLDLR and other signaling partners. ^{9 12 13 14 17}  

CD105 is a homodimeric transmembrane glycoprotein involved in vasculogenesis and angiogenesis. The expression of CD105 can be upregulated in proliferating endothelial cells in vitro and in vivo (for reviews, see Fonsatti et al., ³¹ Duff et al., ³² and Kumar et al. ³³ ). A high level of CD105 expression in HUVECs was strictly associated with cellular activation and an increased proliferation rate. Selective highly intense CD105 staining of vascular endothelial cells was found in tissues undergoing active angiogenesis, whereas none or weak staining of CD105 was detected in blood vessels without angiogenesis. In our study, the weak CD105 signal in the wild-type retina was consistent with quiescent endothelial cells expressing minimal levels of CD105. In contrast, the strong intense CD105 signal in retinal vessels of vldlr ^−/− mice confirmed the activation of endothelial cells in the vldlr ^−/− retina. The upregulation of CD105 was not restricted at the SNV lesion site; rather, it spread throughout the entire retinal vascular network. In addition, we observed strong staining of CD105 antibody in the vldlr ^−/− retina as early as P14, when normal retinal vasculature development was just completed, before retinal neovascularization reached the subretinal space, as we previously described. ²² This widespread and early upregulation of CD105 demonstrated early, global activation of retinal endothelial cells in the vldlr ^−/− retina. 

CD106 is a 110-kDa transmembrane glycoprotein expressed by myeloid lineage and bone marrow stromal cells. It has been reported that endothelial cells constitutively express low levels of CD106 and upregulate the level in response to cytokines, such as TNF-α. ^{34 35} CD106 has been characterized as a marker of endothelial cell activation. However, our study has shown that CD106 can only be detected in large vessels in the GCL without labeling any small vessels in the wild-type inner retina. In contrast, strong CD106 staining was clearly seen at the SNV site of the vldlr ^−/− retina. The differential expression pattern of CD105 and CD106 we observed in the vldlr ^−/− retina clearly revealed that these two markers identified different populations of retinal endothelial cells that overlapped slightly. This indicated that CD105 may be a more sensitive marker for early endothelial cell activation, whereas CD106 labeled normal and pathologic mature endothelial cells in large vessels. Potentially different mechanisms could be involved in different stages of endothelial cell activation. Indeed, studies have shown that activated endothelial cells can express different markers during early and late stages of angiogenesis. ^{36 37} Because endothelial cell activation is known to participate in the initiation, prolongation, and exacerbation of angiogenic processes, clear signs of early endothelial cell activation in the vldlr ^−/− retina strongly suggest that neovascular growth is the consequence of such activation. 

VLDLR in Angiogenesis

The classical view of VLDLR function in lipoprotein transport was challenged when the nonsense mutation of VLDLR in mice exhibited normal plasma lipid and lipoprotein levels with only slightly reduced adipose tissue mass of the mice. ²⁰ Interestingly, subsequent examination revealed a sustained penetration of subretinal neovascular phenotype in the mutant eyes. ¹⁹ This unexpected discovery in the retina without other detectable abnormality indicated a prominent effect of VLDLR on retinal angiogenesis. Previous studies in vascular biology and tumor angiogenesis research have revealed that VLDLR is involved in regulating endothelial cell proliferation and angiogenesis. ^{30 38 39} A recent report also showed increased activity of the wnt pathway in vldlr ^−/− mouse eyes. Downregulation of VLDLR by small interfering RNA resulted in the activation of HUVECs by upregulating wnt signaling molecules, ⁴⁰ suggesting that the wnt signaling pathway plays a role in the negative regulation of VLDLR on angiogenesis. ^{41 42} In the present study, we confirmed the activation of endothelial cells in the vldlr ^−/− retina and demonstrated a direct and broad effect of VLDLR on multiple angiogenic functional properties of retinal endothelial cells. These in vivo and in vitro data support the notion that VLDLR is a critical endogenous molecule maintaining the homeostasis of retinal endothelial cells. Loss of VLDLR leads to disinhibition of retinal endothelial cells, inducing uncontrolled vascular outgrowth and leading to SNV. 

As mentioned, VLDLR has been identified recently as one of the functional candidate genes for a significant association with AMD in humans. ⁸ Microarray screening of hundreds of SNPs in 360 AMD-affected and 360 healthy persons confirmed a positive association of the VLDLR gene with the mixed phenotype of AMD (Sanchez-Salorio M, et al. IOVS 2007;48:ARVO E-Abstract 3007). Although the overall effect of these polymorphisms on VLDLR function is unknown, these findings provide an additional link between the VLDL receptor and abnormal angiogenesis in the retina, implicating an important role for VLDLR in retinal angiogenesis. 

 Supported in part by a postdoctoral fellowship from the Fight for Sight (AJ), research grants from the Reeve’s Foundation and the American Health Assistance Foundation, and an unrestricted grant from Research to Prevent Blindness, Inc.

 Submitted for publication June 18, 2008; revised September 17, 2008; accepted December 26, 2008.

 Disclosure: A. Jiang, None; W. Hu, None; H. Meng, None; H. Gao, None; X. Qiao, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Immunofluorescence staining of CD105 in wild-type (+/+) and vldlr ^−/− (−/−) mouse retina. (A, B) In the retinas of young adult wild-type mice, the signal of CD105 immunoreactivity was minimal. At higher magnification, scattered dim circular, semicircular, and patchy labeling was barely visible in the inner retina. (C) Distribution of the staining pattern corresponded to the retinal vascular structures revealed by isolectin staining in these layers, but only a small fraction of the vascular network was CD105 positive. (D, E) CD105 signal in young adult vldlr ^−/− mice was stronger throughout the entire retina, including the inner retina. The strongest signal was at the neovascular bulb in the subretinal space (arrows). (G, H) An intense CD105 signal was evident early on P14, before the onset of neovascularization in the vldlr ^−/− retina. CD105 staining in the P14 wild-type retina was weak compared with that in the young adult wild-type (data not shown). (F, I) The close match of the staining pattern between CD105 antibody and lectin in the serial sections of the vldlr ^−/− retina indicated that the entire retinal vasculature, not just the lesion site (arrows), was activated.

Figure 2.

 

Retinal sections of young adult wild-type and vldlr ^−/− mice were stained with anti-CD106 antibody. (A) In the wild-type retina, CD106 expression was detected in large vessels in the GCL layer and around the outer limiting membrane. (B) In the vldlr ^−/− retina, in addition to the similar staining pattern, a strong CD106 signal was found in new blood vessels extending from the GCL to the subretinal space (arrow). CD106 signal in the retina was weak at P14 in both wild-type and vldlr ^−/− mice (data not shown).

Figure 3.

 

Growth of primary RVECs isolated from wild-type (+/+) and vldlr ^−/− (−/−) retina in vitro. (A, B) Primary RVEC culture after 5 days in vitro. RVECs from vldlr ^−/− mice reached confluence more quickly than did wild-type RVECs. (C, D) Immunohistochemical staining of RVECs with anti-CD31 (C, red), anti-αSMA (C, green, arrow), and anti-VE-cadherin (D, green) antibodies confirmed over 95% purity of each culture.

Figure 4.

 

In vitro angiogenic assays of RVECs from wild-type (+/+) and vldlr ^−/− (−/−) retina. (A) MTS growth curve assay revealed that vldlr ^−/− RVECs grew consistently more quickly than did wild-type RVECs in the BM (EBM + 0.5% FBS). The difference was statistically significant (n = 4; P < 0.01). The growth of RVECs in both genotypes was further enhanced in the GM, as was the genotypic difference (n = 4; P < 0.001). (B) Tube formation assay of RVECs in the extracellular matrix with basal, bFGF-supplemented, or GM. Quantitation of total tube length in each well revealed striking genotypic differences in all three conditions. The difference became more prominent in 10 ng/mL bFGF or GM (*P < 0.01 compared with wild-type RVECs in the same medium; n = 3). (C) Migration assay of RVECs from both genotypes seeded in transwells with basal, bFGF supplemented, or GM in the lower chamber. Quantitation of the total number of migrating cells revealed significantly more migration of vldlr ^−/− RVECs than that of the wild-type in BM and 10 ng/mL bFGF (*P < 0.01; n = 3). Migration of vldlr ^−/− cells was 1.3 to 3.7 times that of the wild-type.

Figure 5.

 

MTS assay of wild-type RVEC proliferation after VLDLR blocking. (A) Effects of rhRAP, a strong antagonist of VLDLR, on wild-type RVEC growth. (A) Dose-dependent enhancement of wild-type RVEC proliferation was clearly evident after rhRAP treatment. *P < 0.01 (n = 5). Similar boosting effects were observed when wild-type RVECs were treated with rrRAP. (B) Blocking of VLDLR with a cocktail of anti-VLDLR-neutralizing antibodies also promoted the proliferation of wild-type RVECs compared with a control IgG treatment group. *P < 0.01.

The authors thank David Knight for help with manuscript preparation.